This is an extracted Lora (320 Dim) from a classic dreambooth training (800 steps, 16 images, I know it's low but I'm just experimenting there), based on Realistic Vision 2.

The aim of this model was just to test things, but it creates way more accurate Ariana Grande's compared to what's inside the base model !

Original Realistic Vision 2 Ariana Grande ⬇️

✅ BetterAri - Ariana Grande Lora ⬇️

Trigger word : Ariana Grande (of course)

Recommended Negatives: 3D, cgi, red carpet (slight bias on that one)